Mechanisms of Song Perception in Oscine Birds

Brain & Language 115 (2010) 59–68

Contents lists available at ScienceDirect

Brain & Language

journal homepage: www.elsevier.com/locate/b&l

Mechanisms of song perception in oscine birds

Daniel P. Knudsen a, Timothy Q. Gentner b,* a Neurosciences Graduate Program, University of California, San Diego, La Jolla, CA 92093, United States b Department of Psychology, University of California, San Diego, La Jolla, CA 92093, United States article info abstract

Article history: Songbirds share a number of parallels with humans that make them an attractive model system for Accepted 23 September 2009 studying the behavioral and neurobiological mechanisms that underlie the learning and processing of Available online 14 May 2010 vocal communication signals. Here we review the perceptual and cognitive mechanisms of audition in birds, and emphasize the behavioral and neural basis of song recognition. Where appropriate, we point Keywords: out a number of intersections with human vocal communication behavior that suggest common mech- Psychoacoustics anisms amenable to further study, and limitations of birdsong as a model for human language. Auditory Ó 2010 Elsevier Inc. All rights reserved. Vocal communication Language Songbird Evolution Neuroethology Speech

1. Introduction tinuous stream of acoustic information as a collection of behaviorally relevant communication signals and uses this information to Like other communication signals, one adaptive function of affect behavior (Fig. 1). Parallels will be drawn to human vocal birdsong is to influence the behavior of others, usually conspecific communication, and we will present a case for the use of songbirds individuals (Kroodsma & Miller, 1996). Communication signals as a model of certain aspects of human language processing. We achieve this function by transmitting information between the first provide a broad overview of perceptual psychophysics in sender of the signal and the receivers. The success of this transmis- songbirds so that one can appreciate the strong perceptual similar- sion rests on predictability. When a sender produces a specific sig- ities between birds and humans. We then review the behavioral nal, in this case a song, it does so under the expectation that the and neurophysiological work on conspecific song perception, signal will elicit a predictable (i.e. intended) behavior in the recei- focusing on individual vocal recognition mechanisms in European ver. Without the predictable correspondence between production starlings, a species of songbird. and perception, signals would lose their functionality. Thus, the presence of a functional signal implies a reliable correspondence between production and perception mechanisms shaped and 2. Perceptual psychoacoustics maintained by selection pressures. This correspondence confers a special status to communication signals. Like other natural stimuli, It is helpful for any description of a complex system to begin communication signals are often physically complex. Unlike most with a characterization of general abilities along fundamental complex natural stimuli, however, many of the physical dimen- dimensions. Common dimensions along which acoustic signals sions along which communication signals vary can be directly tied are deconstructed are frequency, amplitude, and time, and it is to adaptive behaviors. Research in oscine birds has capitalized on instructive to understand sensory processing at this level. The goal this idea to study the mechanisms underlying the perception and of such research is to inform our understanding of how more cognition of complex natural stimuli (song) in the context of natu- ‘atomic’ descriptions of sounds give rise to the perception of more ral behaviors. complex behaviorally relevant features. Such studies provide an What follows is an account of our current understanding about important context for studies of more complex signals, and estab- the ways in which the songbird auditory system interprets a con- lish the range over which any perceptual ability can operate. The following section provides a brief description of avian psychoacoustics studies supporting the notion that humans and birds * Corresponding author. Address: UCSD, Department of Psychology, 9500 Gilman Dr. MC0109, La Jolla, CA 92093-0109, United States. experience a similar acoustic world. More thorough reviews are E-mail address: [email protected] (T.Q. Gentner). available (Dooling, 1982, 1992, 2000; Fay, 1988).

In addition to hearing single tones, the auditory system must also discriminate between different tones. Understanding the min- imal detectable differences between tones can point to the types of frequency modulations within a signal that are available for a species to use in vocal communication. In general, birds are quite sensitive to changes in frequency, and can discriminate a change in frequency as small as 1% (Dooling, 1982), while humans have even lower detection thresholds across the range of audible frequencies. Birds’ sensitivity to frequency changes also depends on the type of frequency modulation and range of carrier frequencies (Lange- mann & Klump, 1992), results again similar to observations made in human studies (Demany & Semal, 1989; Fastl, 1978). To detect a change in intensity (loudness) between two successive tones, birds require a difference of about 3 dB, humans about 1 dB (Dool- ing, 1982). While humans have quantitatively lower detection thresholds for frequency and loudness discrimination, the similarity of ﬁndings in humans and birds point to auditory systems that are qualitatively similar in the range of psychophysically observa- ble spectral sensitivities.

2.2. Temporal sensitivity

Most acoustic signals unfold over time, and processing in the temporal domain is therefore particularly important. Temporal processing abilities of songbirds have been studied in a variety of ways. A simple measure is the detectability of a sound as its temporal duration increases. In general, the longer the tone is played for, the lower the SPL needed for detection. Consistent with ﬁnd- ings from a host of other animals including humans (see Brown & Maloney, 1986), birds’ thresholds for hearing a pure tone im- prove as the duration is increased from a few milliseconds to 200–300 ms (Dooling, 1980). Another common measure of temporal acuity, known as the gap detection threshold, measures the minimum temporal interval that can be detected between two sounds. Several studies from birds show gap detection thresholds ranging from 2 to 3 ms, which is similar to thresholds found in humans (see Klump & Maier, 1989). This suggests that intervals in natural vocalizations less than 2–3 ms may not be perceived. Dura- tion discrimination measures, which describe an organism’s ability to determine whether one sound has a longer duration than another, are also similar between birds and humans (Maier & Klump, 1990).

2.3. Masking

Fig. 1. Functional auditory behaviors in songbirds. The gray boxes represent the functions that the auditory system must perform to mediate behavior. Text to the In psychoacoustics, the critical ratio describes the ability of an right of these functions gives examples of phenomena observed in songbirds (see organism to perceive a tone in a noisy background. It is defined text). Although this review focuses on songbird auditory processing, many of the as the SPL of a target tone needed for detection divided by the functions of the auditory system are likely conserved across a range of vertebrates, SPL of the background masking noise. Critical ratios are a function including humans. As behavioral complexity increases, so does the likelihood that particular mechanisms are unique to different species. The similarities and the of the frequency of the target tone, and in accordance with previ- differences between species yield powerful comparative hypotheses about the ous studies in humans and in other mammals, critical ratios in behavioral and neural mechanisms for auditory perception and cognition in most songbirds increase at about 3 dB per octave (Dooling et al., vertebrates. 1986; Langemann, Klump, & Dooling, 1995; Okanoya & Dooling, 1987). Most songbirds’ critical ratio curves show a similar shape 2.1. Spectral sensitivity to those of humans and other mammals, though humans show lowered threshold levels on the order of a few decibels (Okanoya Audibility curves, which describe the loudness (sound pressure & Dooling, 1987). level, SPL) required for detectability as a function of frequency, Comodulation masking release (CMR) is a slightly more com- have the same basic shape in many songbird species (Dooling, plex masking phenomenon that has been described in both birds Okanoya, Downing, & Hulse, 1986; Okanoya & Dooling, 1987). The and humans. CMR occurs when sounds that are modulated to- threshold sensitivity for pure tones in most songbird species is best gether across time serve to release each other from masking by around 2–5 kHz, increases gradually as the frequency of the tone overlapping noise. CMR is measured as the effective decrease in becomes lower, and increases quite sharply as the frequency of the threshold SPL afforded by the comodulation. CMR has been pro- tone rises. The typical high-frequency cutoff for songbirds is posed as mechanism for auditory stream segregation (discussed 8–11 kHz. Although audibility thresholds are lower in humans than below), as sounds that are produced from the same source in birds, the overall shape of the audibility curve is similar. take the same path to the listener and are therefore modulated D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68 61 identically. Starlings show robust CMR, and in many ways are sim- that most reliably distinguish their own songs from those of other ilar to humans in terms of amount of masking release and the fre- species in the same sound environment (Nelson & Marler, 1990). quency and bandwidth dependencies of the release (Klump & For example, frequency appears to be the most important cue for Langemann, 1995). species recognition among field sparrows (Nelson, 1989). Recogni- tion systems appear to be tuned to the particular acoustic space of one’s own species’ vocalizations (Nelson, 1988), but also maintain 3. Auditory cognition in birds more general capacities to distinguish among many different types of song. These biases for the discrimination of species-specific Within the broad framework of acoustic processes laid out by vocalizations, and hence mechanisms for vocal recognition, likely the results of psychoacoustic studies, birds (and any other organ- result from evolutionary or ontogenic changes in the (presumably) ism that uses acoustic communication) face a number of specific central processing mechanisms that underlie higher-level auditory challenges. The auditory system must first detect a potentially rel- perception. This is consistent with the more general assumption evant signal against backgrounds of environmental noise or other that the perceptual and cognitive processes underlying individual potentially relevant signals. Signals must be processed to deter- vocal recognition, a widespread phenomenon among songbirds mine their relevance in specific behavioral contexts, and mapped (reviewed by Stoddard (1996)), operate on complex (multi-dimen- onto appropriate behaviors. These behaviors are often mediated sional) acoustic signals. Within the same species many acoustic by learning, memory and attentional mechanisms, implying a suite features may be relevant for classification of song components of cognitive processes that act on complex stimulus representa- (Sturdy, Phillmore, & Weisman, 2000). Recent laboratory studies tions such as stream segregation, discrimination, recognition, cate- of European starlings have addressed these questions directly by gorization, temporal pattern detection, and decision processes. The determining more precisely the form of the acoustic signal control- acoustic diversity of birdsong makes the system attractive for stud- ling recognition in this species. ies of the behavioral and neurobiological bases for many of these cognitive processes. 3.2.1. Individual vocal recognition in starlings Male starlings present their songs in long episodes of continu- 3.1. Stream segregation ous singing referred to as bouts. Song bouts, in turn, are composed of much smaller acoustic units referred to as motifs (Adret-Haus- One challenge to the auditory system is the separation of berger & Jenkins, 1988; Eens, Pinxten, & Verheyen, 1991), which incoming auditory information into behaviorally relevant groups, in turn are composed of still smaller units called notes. Notes can or streams. This is referred to as stream segregation, and under be broadly classified by the presence of continuous energy in their most natural conditions equates to separation of streams based spectrographic representations, and although several notes may on originating source. In order to effectively use the acoustic infor- occur in a given motif, their pattern is usually highly stereotyped mation in one’s environment, it is helpful to group acoustic infor- between successive renditions of the same motif. One can thus mation arising from the same source, and to treat this consider starling song as a sequence of motifs, where each motif information as separate from information arising from other is an acoustically complex event. The number of unique motifs that sources. In human language, for example, this ability allows one a male starling can sing (i.e. his repertoire size) can be quite large, to follow the vocalizations of a speaker across time in a noisy and consequently different song bouts from the same male are not environment. necessarily composed of the same set of motifs. This broad acous- Stream segregation is one of the main tasks of any perceptual tical variation in their song provides several potential cues that system engaged in vocal communication, and it is therefore no sur- starlings might use when learning to recognize the songs of an prise that songbirds are capable of separating one auditory stream individual conspecific and while maintaining that recognition over from another in a variety of conditions. Like humans (Bregman, time. One straightforward recognition mechanism is the associa- 1990), starlings can hear competing streams emerge from a se- tion of specific motifs with specific singers. Although some sharing quence of alternating tones when the frequency difference be- of motifs does occur among captive males (Hausberger, 1997; tween putative ‘‘streams” is large (MacDougall-Shackleton, Hulse, Hausberger & Cousillas, 1995), the motif repertoires of different Gentner, & White, 1998). Likewise, starlings continue to recognize males living in the wild are generally unique (Adret-Hausberger long segments of conspecific songs when noise is added in the form & Jenkins, 1988; Eens, Pinxten, & Verheyen, 1989; Eens et al., of other starling songs, the songs of other species, or an entire 1991). Thus, learning which males sing which motifs can provide ‘dawn chorus’ (Hulse, MacDougall-Shackleton, & Wisniewski, discriminative cues for song recognition. Similar strategies may 1997). Canaries can maintain accurate recognition of conspecific underlie individual song recognition behaviors observed in other song in the face of very loud broadband white noise (Appeltants, species where individual possess elaborate vocal repertoires such Gentner, Hulse, Balthazart, & Ball, 2005). The neural correlates of as song sparrows (Beecher, Campbell, & Burt, 1994; Stoddard, Bee- the so-called cocktail party effect, that is the extraction of a mean- cher, Horning, & Campbell, 1991), but other strategies are available ingful communication signal from noise, have been investigated in as well (Brooks & Falls, 1975a, 1975b; Nelson, 1989). starlings and zebra finches (Bee & Klump, 2004; Narayan et al., Data from operant studies in starlings indicate that song recog- 2007). nition is based at the level of the motif. Starlings trained using operant conditioning to recognize sets of songs from conspecific 3.2. Song discrimination and recognition individuals can readily generalize correct recognition to novel songs from the same singers (Gentner & Hulse, 1998, 2000b). How- Many species of bird are adept at discriminating songs of their ever, recognition falls to chance when these novel song bouts have own species as well as songs from other species. Within this con- no motifs in common with the training songs (Gentner & Hulse, text, birds are generally more adept at discriminating among dif- 2000b). This failure to generalize correct recognition to songs com- ferent conspecific songs than among heterospecific songs (Cynx posed of novel motifs, or to single novel motifs, is consistent with & Nottebohm, 1992; Dooling, Brown, Klump, & Okanoya, 1992; the notion that starlings learn to recognize the songs of individual Sinnott, Sachs, & Hienz, 1980). Fieldwork in sparrows is consistent conspecific singers by attending to information contained at (or be- with these laboratory findings, and suggests that different species low) the level of the motif. That is, they appear to associate distinct may rely, at least for species recognition, on those acoustic features sets of motifs (or variant motif features) with individual singers. 62 D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68

If starlings learn to recognize individuals by the sets of unique of signal elements between individuals increased, the association motifs that they sing, then once learned, it should be possible to of unique vocabularies with different individuals would become control recognition systematically by varying the proportions of increasingly inefficient, forcing the differentiation of vocal identity motifs in a given bout that come from two ‘‘vocally familiar” males. from other semantic channels. That is, recognition behavior ought to follow the proportional dis- tribution of motifs from two vocally familiar males rather than the 3.3. Syntactic pattern sensitivity presence or absence of single diagnostic motifs from either male. The behavioral data confirm this prediction. When starlings are Part of what makes human language unique is its incorporation compelled to classify conspecific songs, they do so by memorizing of long timescale patterning rules (grammars) that constrain the large numbers of unique song components (i.e. motifs) and orga- meaning of specific utterances, e.g. telling us precisely who did nizing subsets of these motifs into separate classes (Gentner & Hul- what to whom. Animal vocalizations lack the combinatorial gram- se, 2000b). Consistent with this idea, recent results demonstrate matical complexity of human language. Yet many animal commu- that motifs form perceptually salient auditory objects embedded nication signals have a rich temporal structure, and understanding in a hierarchy of acoustic patterns (Gentner, 2008). When the how the temporal patterns in such signals are used and perceived sub-motif acoustic structure is permuted in varying ways, recogni- is of broad interest. Such studies can serve to emphasize important tion of the songs is significantly impaired. These effects are only similarities and differences in underlying mechanisms for process- seen for familiar motifs, however, suggesting that starlings learn ing acoustically complex vocal signals. the sub-motif structure of motifs and then perceive them as holis- The prior sections on individual vocal recognition in starlings tic auditory objects (Gentner, 2008). establish the functional importance of motif-level song organiza- Additional support for the notion that motifs form holistic audition. Under normal conditions, however, single motifs are almost tory objects comes from recent results in starlings that mirror per- never produced in isolation but typically occur as part of long ceptual restoration studies in humans (Seeba & Klump, 2009). In and elaborate song bouts where 25–30 different motifs may be the classic demonstration of perceptual restoration (Warren, strung together in close succession. In the following sections we 1970), units of speech (phonemes) within a presented sentence discuss the role of temporal motif patterning in starling song are perceived as present by human observers despite being re- perception. placed by noise (e.g. a human cough). The restoration effect only occurs when the phoneme is replaced by noise; when silence is in- 3.3.1. Motif sequences serted instead of noise, phonemes are correctly perceived as miss- Sensitivity to the temporal patterning of motifs could arise in its ing. This shows that with appropriate contextual cues, observers simplest form based on the explicit sequences in which familiar perceive missing acoustic information at the level of a whole pho- motifs are produced. As an example, a bird might learn to recog- neme, suggesting that they have a memory for phonemes as holis- nize motifs A–D (where letters denote different motifs) and the se- tic auditory units. Similarly, starlings trained to detect small quence ABCD. If this is the case, then presenting the same motifs in changes in motif similarities report significantly greater similarity a different order may affect song recognition. Consistent with this, between intact motifs and those with sections replaced by noise starlings trained to recognize naturally patterned songs have a when compared to motifs with sections replaced by silence. Impor- much harder time recognizing novel songs from familiar singers tantly, these results hold only for familiar motifs, suggesting that when motifs are strung together in a random order, than when mo- memory for motifs as learned objects informs their perception at tifs follow each singer’s natural motif ordering (Gentner & Hulse, later times (Seeba & Klump, 2009). These results provide another 1998), indicating that starlings do attend to motif sequencing. line of evidence for considering motifs as acoustic units and suggest that the representation of these motifs emerges via a combi- 3.3.2. Motif patterns nation of feed forward auditory input and memory for the motif Additional studies of temporal pattern perception in starlings as an object. move beyond simple transition probabilities between adjacent As a cognitive recognition strategy, classifying songs according motifs to ask if starlings can acquire abstract rules that describe to their component (motif) structure represents a straightforward the patterning of familiar motifs (Gentner, Fenn, Margoliash, & method of dealing with these complex acoustic signals. Because Nusbaum, 2006). That is, instead of learning the explicit sequence individual starlings tend to possess unique motif repertoires, dis- of motifs ABCD as above, the study asked if starlings could learn joint sets of motifs will generally correspond to individual identity. the pattern ABCD, where the letters now denote sets of motifs that Therefore, attending to the motif structure captures a significant can occur at each position. Rules that describe sequences of pat- portion of the individual variation in the signal, albeit at the ex- terned strings have a rich history in the theory of formal grammars, pense of a heavy demand on memory capacity. From a human per- and form the basis for theories of syntax in human language. For- spective, this might seem a suboptimal strategy for individual mal grammars can be classified hierarchically according to the vocal recognition, because in humans the total set of vocal signals complexity of the patterns they can produce or recognize (Hopcroft that can be produced by a single individual is theoretically infinite. & Ullman, 1979). Finite-state grammars are among the most Instead, individual talker recognition in humans relies on voice limited type of formal grammar, and have been thought to describe characteristics such as timbre, glottal pulsation frequency, and all animal communication systems (Hauser, Chomsky, & Fitch, spectral contours imparted by laryngeal morphology (Bricker & 2002; but see Suzuki, Buck, & Tyack, 2006). Human languages Pruzansky, 1976) that are largely independent of the linguistic con- minimally require a grammar more complex than finite-state, tent (Remez, Fellowes, & Rubin, 1997). Although recent results called a context-free grammar, in part to support the hierarchical indicate that some voice characteristics are present in starling song embedding common in many syntactic structures (Chomsky, (Gentner, 2008), these birds appear to recognize individuals by 1957; Hopcroft & Ullman, 1979). preferentially using information at longer temporal scales rather Starlings can learn to classify patterned strings of song motifs than the short (fast) timescale features that carry voice character- generated by both a finite-state grammar and a context-free gram- istics. This observation is consistent with the conjecture that the mar (Gentner et al., 2006). In this case, strings generated by the incorporation of individual identity information into the phonetics context-free grammar followed the form A2B2, while the strings of the signal may be a plesiomorphic (evolutionarily older) state generated by the finite-state grammar followed the form (AB)2, for vocal communication. As the size of the lexicon and sharing where A and B refer to sets of acoustically distinct starling song D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68 63 motifs known as ‘‘rattles” and ‘‘warbles”. It is important to note that the context-free grammar learned in this study, while more computationally complex than the finite-state grammar, is not sufficient to fully capture the hierarchical structure in human languages. Specifically, it lacks any long-range dependencies between specific elements that, in for example some English sen- tences, help define the pairings between subjects and verbs, and thereby enable understanding of who did what to whom. There- fore, one should not conclude from this study that songbirds have the capacity to master ‘‘human” grammar. Nonetheless, it is clear that they can extract patterning rules from strings of vocal signals, and that those rules can have a remarkable level of complexity. Whether or not birds and humans rely on similar neuronal mechanisms to extract temporal pattern rules is unknown. In any case, the basic ability to attend to high-level patterns within strings of vocal signals, including at least some patterns defined by formal grammars, does not appear to be a recent or unique adaptation. Although there has been some discussion of the role of syntactic rules in structuring bird song in chickadees and wrens (Clucas, Freeberg, & Lucas, 2004; Hailman & Ficken, 1986; Holland, Dabels- teen, & Paris, 2000), the existence of complex pattering rules in songbird vocalizations has not received adequate attention. Cur- rently, there is no strong evidence to support the notion that songbirds (or any non-human species) use syntax to vary the semantic content of vocal signals in the combinatorial manner. By itself, the ability to process simple syntactic structures may be of little functional significance or adaptive value, and may represent a necessary but insufficient precondition to the use of unbounded signal sets observed in humans. These sorts of primitive pattern recognition abilities might need to be paired with more sophisticated ges- tural systems as a permissive step to subsequent evolution of human language processing capacities in their modern forms.

4. The neural coding in the songbird auditory system

The behavioral data yield several hypotheses regarding the neural mechanisms underlying the recognition of natural (i.e. complex) acoustic signals. First, the functionality of auditory objects in recognition behavior implies an explicit and discrete representation for vocal signals in the central nervous system. That is, the response functions of single neurons or of populations of neurons in appropriate auditory regions should reflect the segmentation of song at the level of the motif (in starlings) or other behaviorally relevant temporal scales. Second, because vocal recognition behavior requires the learned association between songs and singers, the Fig. 2. Schematic of the major auditory processing regions in the avian brain. Auditory information projects from the ear to the brainstem cochlear nuclei. In neural representations of motifs should reflect their explicit behav- general, more complex acoustic information is represented in more central ioral relevance. That is, there should be a bias for representations of structures, i.e. those further from the cochlea. Apart from those to the classic familiar compared to unfamiliar motifs. Third, the representational song-control system, projections from the primary auditory pathways to areas mechanisms and capacity (i.e. memory) of the system should per- responsible for multimodal integration and non-singing behaviors are uncharac- mit the acquisition of very large numbers of acoustically complex, terized, but presumably exist. See the text for more details. Thinner arrows indicate direct neural projections. Abbreviations: CM: caudal mesopallium, MLd: mesen- natural objects (motifs). The nature of object representation, repre- cephalicus lateralis dorsalis, NCM: caudomedial nidopallium, Ov: nucleus ovoidalis. sentational plasticity, and memory capacity are central questions for any researcher interested in the neural coding of natural stimuli, including human speech and language. Below, we consider tude peaks in the sound pressure waveform of the stimulus (Sachs these questions in the context of the neural representation of star- & Sinnott, 1978). Afferents from the brainstem converge in the ling song. midbrain structure mesencephalicus lateralis dorsalis (MLd). MLd is anatomically analogous to the mammalian inferior colliculus 4.1. The avian auditory system (IC), and in many respects it is also functionally analogous. Like the mammalian IC, MLd neurons show a wide variety of tuning Auditory information enters the avian brain via sensory trans- characteristics, are not particularly sensitive to level differences, duction in the inner ear, where vibrations on the basilar membrane and have temporally precise responses that preserve the timing of the cochlea are transduced into the action potentials propagated information in the acoustic signal (Woolley & Casseday, 2004). to the cochlear nuclei in the brainstem (Fig. 2). The cochlear nuclei They also respond to white noise and conspecific song in a way are characterized by strong tonotopy, and the neurons within these that suggests selectivity for particular spectrotemporal features, nuclei have a tendency to fire tonic bursts that align with ampli- and the vast majority of these cells follow sinusoidal amplitude 64 D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68 modulations – suggesting again that these cells encode the tempo- maintenance of a bird’s own song. While the full range of efferent ral features of an acoustic signal with high fidelity (Woolley, Fre- targets for the auditory telencephalon is unknown, it is likely that mouw, Hsu, & Theunissen, 2005). brain regions outside the classic ‘‘song system” are involved in the MLd sends afferents into the avian auditory thalamus, the Nu- subsequent processing of song representations, including areas cleus Ovoidalis (Ov). Ov shows a strong tonotopic arrangement, responsible for integrating multimodal information that naturally and its neurons have high spontaneous firing rates, and respond accompanies hearing song, as well as regions necessary for carry- to stimuli at their preferred frequency with a tonic increase in firing out decision processes and motor programs necessary for exe- ing rate (Bigalke-Kunz, Rübsamen, & Dörrscheidt, 1987). There is cuting song-mediated behaviors, e.g. copulation displays. also evidence of lateral inhibition in this region, as many cells show An increasing amount is known about the functionality and suppressive sidebands in their frequency tuning curves (Bigalke- gross anatomical connectivity of the structures described above, Kunz et al., 1987). Thalamic efferents project from Ov mainly into but there is still little knowledge of the underlying functional mi- the Field L complex (shortened here to Field L), the avian analog of cro-circuitry. Early EM and light microscopy studies of local cir- the mammalian primary auditory cortex (Karten, 1968, 1969). Spe- cuitry within Field L indicate that neurons are arranged in small cifics of cytoarchitecture and neural connectivity allow further clusters (Saini & Leppelsack, 1977), and it is tempting to hypothe- subdivision of the Field L complex into regions L1, L2a, L2b, L3 size that these clusters may form functional units. Perhaps related and L (here ‘L’ refers to a subregion of the larger complex Fortune to this, large numbers of GABAergic cells are distributed through- & Margoliash, 1992; Saini & Leppelsack, 1977, 1981). L2a is the pri- out Field L, NCM, and CMM. In NCM and CMM many of these cells mary thalamorecipient zone, but L1 and L3 also receive weak tha- also show song-inducible expression of ZENK (Pinaud et al., 2004), lamic input (Vates, Broome, Mello, & Nottebohm, 1996). Field L suggesting that inhibitory mechanisms play an important func- shows strong tonotopy and robust responses to both natural song tional role in the representation of learned acoustic material (see and pure tones (Capsius & Leppelsack, 1996, 1999; Haüsler, 1996; the following for further discussions of inhibitory processing in Leppelsack & Vogt, 1976; Sen, Theunissen, & Doupe, 2001). the avian auditory forebrain Capsius & Leppelsack, 1996; Pinaud From Field L, forebrain processing proceeds along a number of & Terleph, 2008; Pinaud et al., 2008). Much work remains to be parallel and interconnected pathways. Field L itself projects to done to test these hypotheses. In general, elucidating the structure the caudomedial nidopallium (NCM), which surrounds Field L ana- and activity of local circuitry will be crucial for understanding how tomically, and to the lateral portion of the caudal mesopallium song representation differs in distinct brain regions, the transfor- (CLM) (Vates et al., 1996). The medial portion of the caudal mes- mations that occur between these regions, and the role of these dif- opallium (CMM) shares reciprocal projections with both CLM and ferent representations in a bird’s behavior. This will be an area of NCM (Vates et al., 1996). Neurons throughout the songbird audi- considerable research effort over the next decade. tory telencephalon show complex patterns of tonotopic organiza- The avian forebrain lacks the gross anatomical structure of tion (Capsius & Leppelsack, 1996; Haüsler, 1996; Leppelsack & mammalian cortex. Nonetheless, birds and mammals use their Schwartzkopff, 1972; Rübsamen & Dörrscheidt, 1986), and varying auditory systems to solve many similar behavioral problems. selectivity to species-specific vocalizations (Bonke, Bonke, & Structural and functional similarities present in the auditory re- Scheich, 1979; Leppelsack & Vogt, 1976; Müller & Scheich, 1985). gions of both classes of animals at the cellular (Saini & Leppelsack, Neurons in Field L2a/b are less selective for species-specific vocal- 1981) and systems levels (Wild, Karten, & Frost, 1993), (Sockman, izations than those in L1 and L3 (Theunissen & Doupe, 1998; The- Gentner, & Ball, 2002), and across the whole of the forebrain (Jarvis unissen, Sen, & Doupe, 2000; Theunissen et al., 2004), which in et al., 2005) may reflect these common behavioral goals, Observa- turn are less selective than those in NCM and CM (Grace, Amin, tions suggesting a link between avian studies and the rich body of Singh, & Theunissen, 2003; Müller & Scheich, 1985). Increasing research on the mammalian cortex highlight the need for compar- song selectivity across these regions is coincident with an increase ative studies of auditory system function. Comparing anatomical in the non-linear components of the neural responses (Sen et al., and functional similarities between birds and mammals and across 2001), and higher regions in the pathway (i.e. NCM and CMM) species within these classes will lend insight into which processes are involved in the representation of complex acoustic features are likely to be fundamental to the representation of the auditory (e.g. Bonke et al., 1979; Gentner, Hulse, & Ball, 2004; Gentner & world by a neural substrate, and which may be idiosyncratic to a Margoliash, 2003; Leppelsack & Vogt, 1976; Müller & Scheich, particular organism or group of organisms. 1985). Additional support for the role of NCM and CM in the processing 4.2. Experience-dependent representation of song of conspecific song comes from studies of stimulus driven expression of the immediate-early-gene (IEG) ZENK, a commonly used The acoustic environment of a songbird is not likely to be con- marker of song-evoked auditory activity (Mello, Velho, & Pinaud, stant over time, or between individuals. Territories change hands 2004a) and song-induced experience-dependent plasticity (Jarvis, and the available range of mates fluctuates. In order to effectively Mello, & Nottebohm, 1995; Jones et al., 2001; Mello, Nottebohm, interpret the acoustic features in song, the auditory system needs & Clayton, 1995; Mello, Velho, & Pinaud, 2004b; Mello, Vicario, & to structure its representation in such a way that allows for dy- Clayton, 1992; Ribeiro, Cecchi, Magnasco, & Mello, 1998). In star- namic behavioral goals of the organism. A combination of gene lings, the IEG response in NCM appears tied to stimulus novelty, expression and electrophysiological studies has helped to charac- whereas IEG activity in CMM appears to correlate with the ongoing terize several regions within the auditory telencephalon where recognition of familiar songs (Gentner et al., 2004). Together, these representations of learned song are thought to be stored. observations are consistent with a functional hierarchical organi- Many neurons in NCM and CM show a rapid and selective up- zation of the songbird auditory system that is tuned throughout regulation of the ZENK (or its protein product) in response to the to conspecific song (Hsu, Woolley, Fremouw, & Theunissen, 2004; presentation of conspecific songs (Mello et al., 1992), in manner Woolley, Gill, Fremouw, & Theunissen, 2009; Woolley et al., 2005). sensitive to the acoustics of particular song syllables (reviewed in In addition to connections between regions within the auditory Pinaud & Terleph, 2008; Ribeiro et al., 1998; Terleph, Mello, & Vica- system, anatomical (Vates et al., 1996) and functional (Bauer et al., rio, 2006). Interestingly, these IEG responses (Mello et al., 1995), as 2008; Shaevitz & Theunissen, 2007) connections have been ob- well as NCM multiunit recordings (Chew, Vicario, & Nottebohm, served between auditory processing areas and the song production 1996), habituate to the repeated presentation of the same conspe- system, providing a path for feedback during the acquisition and cific song. Similar habituation in the ZENK response in NCM also D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68 65 appears closely tied to the context under which song is presented The observations of experience-dependent plasticity in the (Kruse, Stripling, & Clayton, 2004). In addition, IEG responses are songbird auditory system suggest a system that is simultaneously elevated during specific components of vocal recognition (Gentner constrained in its immediate representational capacities by each et al., 2004), female choice (Gentner, Hulse, Duffy, & Ball, 2001), animal’s history, yet is tremendously adaptive in its ability to ac- and other song-based learned behaviors (Jarvis et al., 1995) includ- quire a broad (currently undetermined) range of complex repre- ing vocal acquisition (Phan, Pytte, & Vicario, 2006). While these sentations. It is tempting to speculate that similar experience- habituation processes are likely to be important for understanding dependent, hierarchical processes may give rise to the phonetic, how the system processes conspecific song (e.g. Dong & Clayton, phonemic, syllabic and word-level representations that must be 2009), non-associative forms of learning are insufficient to capture correlated with language experience in humans. At the same time, the complexity of most song-based natural behaviors where rein- one must be careful to temper these kinds of speculations with an forcement and stimulus associations are vital. appropriate appreciation for the natural history of the organism. An organism’s sensory experiences evoke long-lasting changes While the song recognition system expresses tremendous plastic- that bias cortical circuits towards representations of behaviorally ity, it is reasonable to ask whether all of the links between auditory relevant signals (Gilbert, Sigman, & Crist, 2001). Experience-depen- features and the behaviorally relevant phenomena they represent dent plasticity in neural responses underlies many forms of asso- must be learned during an individual bird’s lifetime. The existence ciative learning and has been observed across a variety of of, for example, species-specific warning calls and the preference vertebrate sensory systems and brain regions (e.g. Bakin & Wein- for conspecific song suggest that the meaning associated with berger, 1990; Kay & Laurent, 1999; Kilgard, 2003). Experience some signals may not require any sort of learning. By extension, dependant plasticity has also been observed in starlings that have such processes may entail substantially different mechanisms at been trained, using methods similar to those described above, to both neural and behavioral levels. recognize sets of conspecific songs. In birds trained this way, single In many songbirds, the females’ decision of who to mate with neurons in CMM respond much more robustly (i.e. emit many (termed female choice) is driven by the acoustics of competing more action potentials) to songs the birds have learned to recog- males’ songs. (Catchpole & Slater, 1995; Searcy, 1992). These nize than to unfamiliar conspecific songs. The strong response bias behaviors provide an interesting context within which to explore for familiar songs is consistent in animals trained with either a the relationship between learned and innate signal processing two-alternative choice procedure, where all of the training songs mechanisms, as they are strongly influenced by both kinds of infor- are paired with positive reinforcement, or go/no-go procedure, mation. Untrained female starlings show strong preferences for the where only half the training songs are paired with positive rein- long songs of older males (Gentner & Hulse, 2000a), and these pref- forcement. In the latter case, however, songs associated with posi- erences are reflected in ZENK protein expression levels in NCM tive reinforcement elicited significantly stronger responses from (Gentner et al., 2001; see also Maney, MacDougall-Shackleton, CMM neurons than those associated with no reinforcement (Gent- MacDougall-Shackleton, Ball, & Hahn, 2003). Yet, the magnitude ner & Margoliash, 2003). Thus, plasticity in CMM neuronal re- of these preferences can be modified with relatively little experi- sponses cannot be driven simply by song exposure. Other ence (Sockman, Gentner, & Ball, 2005; Sockman et al., 2002). Le- processes that mark the relevance of each song independent of sions to CM, the same region involved in song recognition in any particular signal acoustics, such as attention, motivation and/ starlings, lead female zebra finches to produce copulation solicita- or specific reinforcement, are likely involved. tions to both conspecific and heterospecific song (Macdougall- CMM neurons do not respond at high rates to the entire ensem- Shackleton et al., 1998). Similar effects are reported in female ble of familiar songs. Instead, many individual neurons are selec- canaries with lesions to the song nucleus HVC (Brenowitz, 1991), tive for only a single song, or for only a small number of motifs suggesting that brain regions involved in song production may also within a single song (Gentner & Margoliash, 2003). Thus the same function in song perception. Future work should continue to target features that drive individual vocal recognition of starling song at the neural mechanisms that support this natural decision behavior. the behavioral level (i.e. motifs) also elicit selective responses from single neurons. The mechanisms whereby this selectivity arises over the course of learning are not clear. It may be that the recep- 5. Conclusions tive fields of individual neurons are modified to match the spectrotemporal acoustics of behaviorally relevant signals. Rapid shifts in Although no two species are identical, this review describes a auditory receptive fields driven by behavioral goals have been ob- number of ways in which the auditory processing capacities of served in other systems (Fritz, Elhilali, & Shamma, 2007; Fritz, songbirds are similar to those of other vertebrates, and in particu- Shamma, Elhilali, & Klein, 2003). Alternatively, neurons that best lar to those of humans. Perhaps one reason why birdsong has cap- match the acoustics on relevant songs may be somehow selected tivated humans for so many centuries is that we ‘‘hear the world” from an otherwise static population, and/or the responses neurons in very similar ways. Indeed, the fact that birds and humans share a that don’t match the relevant acoustics suppressed. These possibil- similar perceptual representation of the acoustic world holds the ities, along with the larger question of how to best model the promise that higher-level auditory features relevant to more com- receptive fields of these high-level auditory neurons, remain the plex behaviors in birds may inhabit a similar acoustic space to topic of future research (e.g. Gill, Woolley, Fremouw, & Theunissen, those that are relevant for vocal communication in humans. In 2008). Nonetheless, recent work consistent with the ideas above turn, there may be common neural and behavioral mechanisms suggests that there may be a variety of forms of experience-depen- adaptive for processing signals within this shared acoustic space. dent plasticity throughout the songbird auditory forebrain (Capo- After human language, birdsong is arguably the most acoustically rello & Gentner, 2008; Jeanne, Sharpee, & Gentner, 2008; complex and diverse communication signal known. Although it is Thompson & Gentner, 2008). Spike trains in Field L already carry interesting in it own right, the parallels to human perception make sufficient information to discriminate between conspecific songs it an excellent model for investigations for higher-level auditory (Narayan, Graña, & Sen, 2006), suggesting that representation of processes, and enable a full range of neurobiological and behav- the relevant acoustics may not be the primary role for processing ioral techniques that are difficult to use in humans. within these regions. Rather, the system may be primarily con- Although it is likely that the bird’s brain can help us understand cerned with the extraction of task-relevant features and their map- important strategies for solving auditory perception tasks that ping onto appropriate responses. are mutual between humans and birds, birdsong is not language. 66 D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68

In particular, birdsong lacks the rich combinatorial structure and Catchpole, C. K., & Slater, P. B. J. (1995). Bird song. Cambridge: Cambridge University semantics of human vocal signals, and may be more analogous to Press. Chew, S. J., Vicario, D. S., & Nottebohm, F. (1996). A large-capacity memory system speech than language (Doupe & Kuhl, 1999). By some theories, that recognizes the calls and songs of conspecifics. Proceedings of the National much of what makes human language unique is modality indepen- Academy of Science, 93, 1950–1955. dent (e.g. Hauser & Bever, 2008), and so the greatest differences be- Chomsky, N. (1957). Syntactic structures. The Hague: Mouton. Clucas, B. A., Freeberg, T. M., & Lucas, J. R. (2004). Chick-a-dee call syntax, social tween humans and other animals may lie in the mapping between context, and season affect vocal responses of Carolina chickadees (Poecile the outputs of speech processing brain areas and higher-order carolinensis). Behavioral Ecology and Sociobiology, 57, 187–196. structures. The mechanisms for extracting, parsing, and represent- Cynx, J., & Nottebohm, F. (1992). Testosterone facilitates some conspecific song discriminations in castrated zebra finches (Taeniopygia guttata). Proceedings of ing patterned strings of vocal signals may be conserved across a the National Academy of Sciences of the United States of America, 89(4), wide range of species. Within this context, there remain a number 1376–1378. of substantial challenges for the songbird auditory system that are Demany, L., & Semal, C. (1989). Detection thresholds for sinusoidal frequency modulation. Journal of the Acoustical Society of America, 85(3), 1295–1301. germane to our general understanding of systems neuroscience. Dong, S., & Clayton, D. F. (2009). Habituation in songbirds. Neurobiology of Learning How are the complete receptive fields for high-level auditory (or and Memory, 92(2), 183–188. other sensory modality) neurons best modeled? The best current Dooling, R. J. (1980). Behavior and psychophysics of hearing in birds. In A. N. Popper models perform poorly beyond the primary thalamorecipient & R. R. Fay (Eds.), Comparative studies of hearing in vertebrates. New York: Springer-Verlag. zones. What is the micro-circuitry of Field L, NCM, CM? The orga- Dooling, R. J. (1982). Auditory perception in birds. Acoustic communication in birds nization of the avian auditory telencephalon may be substantially (Vol. 1). Academic Press. different from that of mammalian cortex, yet very similar function Dooling, R. J. (1992). Hearing in birds. In D. Webster, R. Fay, & A. N. Popper (Eds.), The Evolutionary Biology of Hearing (pp. 545–559). New York: Springer-Verlag. is achieved. How? What are the sources – behavioral, anatomical Dooling, R. J. (2000). Hearing in birds and reptiles. In R. J. Dooling, R. R. Fay, & A. N. and physiological – for top-down modulation through attention, Popper (Eds.), Comparative hearing: Birds and reptiles (pp. 308–359). New York: reinforcement, memory, etc. within the system? Finally, what are Springer-Verlag. Dooling, R. J., Brown, S. D., Klump, G. M., & Okanoya, K. (1992). Auditory perception the roles of judgment and decision in the auditory telencephalon of conspecific and heterospecific vocalizations in birds: Evidence for special and its efferent targets? Further studies into the behavioral and processes. Journal of Comparative Psychology, 106(1), 20–28. neural basis of the birdsong auditory system promise to be instru- Dooling, R. J., Okanoya, K., Downing, J., & Hulse, S. H. (1986). Hearing in the starling (Sturnus vulgaris): Absolute thresholds and critical ratios. Bulletin of the mental in answering these questions. Psychonomic Society, 24(6), 462–464. Doupe, A. J., & Kuhl, P. K. (1999). Birdsong and human speech: Common themes and mechanisms. Annual Review of Neuroscience, 22, 567–631. References Eens, M., Pinxten, M., & Verheyen, R. F. (1989). Temporal and sequential organization of song bouts in the European starling. Ardea, 77, 75–86. Adret-Hausberger, M., & Jenkins, P. F. (1988). Complex organization of the warbling Eens, M., Pinxten, M., & Verheyen, R. F. (1991). Male song as a cue for mate choice in song in starlings. Behaviour, 107, 138–156. the European starling Sturnus vulgaris. Behaviour, 116(3–4), 210–238. Appeltants, D., Gentner, T. Q., Hulse, S. H., Balthazart, J., & Ball, G. F. (2005). The Fastl, H. (1978). Frequency discrimination for pulsed versus modulated tones. effect of auditory distractors on song discrimination in male canaries (Serinus Journal of the Acoustical Society of America, 63(1), 275–277. canaria). Behavioural Processes, 69(3), 331–341. Fay, R. R. (1988). Hearing in vertebrates: A psychophysics databook. Winnetka: Hill – Bakin, J. S., & Weinberger, N. M. (1990). Classical conditioning induces CS-specific Fay Associates. receptive field plasticity in the auditory cortex of the guinea pig. Brain Research, Fortune, E. S., & Margoliash, D. (1992). Cytoarchitectonic organization and 536(1–2), 271–286. morphology of cells of the field L complex in male zebra finches (Taenopygia Bauer, E. E., Coleman, M. J., Roberts, T. F., Roy, A., Prather, J. F., & Mooney, R. (2008). A guttata). Journal of Comparative Neurology, 325(3), 388–404. synaptic basis for auditory-vocal integration in the songbird. Journal of Fritz, J. B., Elhilali, M., & Shamma, S. A. (2007). Adaptive changes in cortical receptive Neuroscience, 28(6), 1509–1522. fields induced by attention to complex sounds. Journal of Neurophysiology, 98(4), Bee, M., & Klump, G. M. (2004). Primitive auditory stream segregation: A 2337–2346. neurophysiological study in the songbird forebrain. Journal of Neurophysiology, Fritz, J., Shamma, S., Elhilali, M., & Klein, D. (2003). Rapid task-related plasticity of 92(2), 1088–1104. spectrotemporal receptive fields in primary auditory cortex. Nature Beecher, M., Campbell, S. E., & Burt, J. (1994). Song perception in the song sparrow: Neuroscience, 6(11), 1216–1223. Birds classify by song type but not by singer. Animal Behaviour, 47, 1343–1351. Gentner, T. Q. (2008). Temporal scales of auditory objects underlying birdsong vocal Bigalke-Kunz, B., Rübsamen, R., & Dörrscheidt, G. J. (1987). Tonotopic organization recognition. Journal of the Acoustical Society of America, 124(2), 1350–1359. and functional characterization of the auditory thalamus in a songbird, the Gentner, T. Q., Fenn, K. M., Margoliash, D., & Nusbaum, H. C. (2006). Recursive European starling. Journal of Comparative Physiology A: Neuroethology, Sensory, syntactic pattern learning by songbirds. Nature, 440(7088), 1204–1207. Neural, and Behavioral Physiology, 161(2), 255–265. Gentner, T. Q., & Hulse, S. H. (1998). Perceptual mechanisms for individual vocal Bonke, B. A., Bonke, D., & Scheich, H. (1979). Connectivity of the auditory forebrain recognition in European starlings, Sturnus vulgaris. Animal Behaviour, 56(3), nuclei in the guinea fowl (Numida meleagris). Cell and Tissue Research, 200(1), 579–594. 101–121. Gentner, T. Q., & Hulse, S. H. (2000a). Female European starling preference and Bregman, A. S. (1990). The auditory scene auditory science analysis: The perceptual choice for variation in conspecific male song. Animal Behaviour, 59(2), 443–458. organization of sound. Cambridge: MIT Press. Gentner, T. Q., & Hulse, S. H. (2000b). Perceptual classification based on the Brenowitz, E. A. (1991). Altered perception of species-specific song by female birds component structure of song in European starlings. Journal of the Acoustical after lesions of a forebrain nucleus. Science, 251(4991), 303–305. Society of America, 107(6), 3369–3381. Bricker, P. D., & Pruzansky, S. (1976). Speaker recognition. In N. J. Lass (Ed.), Gentner, T. Q., Hulse, S. H., & Ball, G. F. (2004). Functional differences in forebrain Contemporary issues in experimental phonetics (pp. 295–326). New York: auditory regions during learned vocal recognition in songbirds. Journal of Academic Press. Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Brooks, R. J., & Falls, J. B. (1975a). Individual recognition by song in white-throated Physiology, 190(12), 1001–1010. sparrows. I. Discrimination of songs of neighbors and strangers. Canadian Gentner, T. Q., Hulse, S. H., Duffy, D., & Ball, G. F. (2001). Response biases in auditory Journal of Zoology, 53, 879–888. forebrain regions of female songbirds following exposure to sexually relevant Brooks, R. J., & Falls, J. B. (1975b). Individual recognition by song in white-throated variation in male song. Journal of Neurobiology, 46(1), 48–58. sparrows. III. Song features used in individual recognition. Canadian Journal of Gentner, T. Q., & Margoliash, D. (2003). Neuronal populations and single cells Zoology, 53, 1749–1761. representing learned auditory objects. Nature, 424(6949), 669–674. Brown, C. H., & Maloney, C. G. (1986). Temporal integration in two species of old Gilbert, C. D., Sigman, M., & Crist, R. E. (2001). The neural basis of perceptual world monkeys: Blue monkeys (Cercopithecus mitis) and grey-cheeked learning. Neuron, 31(5), 681–697. mangabeys (Cercocebus albigena). The Journal of the Acoustical Society of Gill, P., Woolley, S. M. N., Fremouw, T. E., & Theunissen, F. E. (2008). What’s that America, 79(4), 1058–1064. sound? Auditory area CLM encodes stimulus surprise, not intensity or intensity Caporello, E., Gentner, T. Q. (2008). Song recognition learning drives experience- changes. Journal of Neurophysiology, 99(6), 2809–2820. dependent changes in the avian telencephalic auditory area field L. Paper presented Grace, J. A., Amin, N., Singh, N. C., & Theunissen, F. E. (2003). Selectivity for at the society for neuroscience. conspecific song in the zebra finch auditory forebrain. Journal of Capsius, B., & Leppelsack, H. J. (1996). Influence of urethane anesthesia on neural Neurophysiology, 89(1), 472–487. processing in the auditory cortex analogue of a songbird. Hearing Research, Hailman, J. P., & Ficken, M. S. (1986). Combinational animal communication with 96(1–2), 59–70. computable syntax: Chick-a-dee calling qualifies as ‘language’ by structural Capsius, B., & Leppelsack, H. (1999). Response patterns and their relationship to linguistics. Animal Behaviour, 34, 1899–1901. frequency analysis in auditory forebrain centers of a songbird. Hearing Research, Hausberger, M. (1997). Social influences on song acquisition and sharing in the 136(1–2), 91–99. European starling (Sturnus vulgaris). In C. Snowden & M. Hausberger (Eds.), D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68 67

Social influences on vocal development (pp. 128–156). Cambridge: Cambridge Mello, C. V., Vicario, D. S., & Clayton, D. F. (1992). Song presentation induces gene- University Press. expression in the songbird forebrain. Proceedings of the National Academy of Hausberger, M., & Cousillas, H. (1995). Categorization in birdsong: From Sciences of the United States of America, 89(15), 6818–6822. behavioural to neuronal responses. Behavioural Processes, 35(1–3), 83–91. Müller, S. C., & Scheich, H. (1985). Functional organization of the avian auditory field Hauser, M. D., & Bever, T. (2008). Behavior: A biolinguistic agenda. Science, L. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and 322(5904), 1057–1059. Behavioral Physiology, 156(1), 1–12. Haüsler, U. (Ed.). (1996). Measurement of short-time spatial activity patterns during Narayan, R., Best, V., Ozmeral, E., McClaine, E., Dent, M., Shinn-Cunningham, B., et al. auditory stimulation in the starling. New York, NY: Plenum Press. (2007). Cortical interference effects in the cocktail party problem. Nature Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, Neuroscience, 10(12), 1601–1607. who has it, and how did it evolve? Science, 298(5598), 1569–1579. Narayan, R., Graña, G., & Sen, K. (2006). Distinct time scales in cortical Holland, J., Dabelsteen, T., & Paris, A. L. (2000). Coding in the song of the wren: discrimination of natural sounds in songbirds. Journal of Neurophysiology, Importance of rhythmicity, syntax and element structure. Animal Behaviour, 96(1), 252–258. 60(4), 463–470. Nelson, D. A. (1988). Feature weighting in species song recognition by the field Hopcroft, J., & Ullman, J. (1979). Introduction to automata theory, languages, and sparrow (Spizella pusilla). Behaviour, 106(1–2), 158–181. computation. Reading, MA: Addison-Wesley. Nelson, D. A. (1989). Song frequency as a cue for recognition of species and Hsu, A., Woolley, S. M., Fremouw, T. E., & Theunissen, F. E. (2004). Modulation power individuals in the field sparrow (Spizella pusilla). Journal of Comparative and phase spectrum of natural sounds enhance neural encoding performed by Psychology, 103(2), 171–176. single auditory neurons. Journal of Neuroscience, 24(41), 9201–9211. Nelson, D. A., & Marler, P. (1990). The perception of birdsong and an ecological Hulse, S. H., MacDougall-Shackleton, S. A., & Wisniewski, A. B. (1997). Auditory concept of signal space. In W. C. Stebbins & M. A. Berkely (Eds.). Comparative scene analysis by songbirds: Stream segregation of birdsong by European perception: Complex signals (Vol. 2, pp. 443–477). New York: John Wiley and starlings (Sturnus vulgaris). Journal of Comparative Psychology, 111(1), 3–13. Sons. Jarvis, E. D., Gunturkun, O., Bruce, L., Csillag, A., Karten, H., Kuenzel, W., et al. (2005). Okanoya, K., & Dooling, R. J. (1987). Hearing in passerine and psittacine birds: A Avian brains and a new understanding of vertebrate brain evolution. Nature comparative study of absolute and masked auditory thresholds. Journal of Reviews Neuroscience, 6(2), 151–159. Comparative Psychology, 101(1), 7–15. Jarvis, E. D., Mello, C. V., & Nottebohm, F. (1995). Associative learning and stimulus Phan, M. L., Pytte, C. L., & Vicario, D. S. (2006). Early auditory experience generates novelty influence the song-induced expression of an immediate early gene in long-lasting memories that may subserve vocal learning in songbirds. the canary forebrain. Learning and Memory, 2, 62–80. Proceedings of the National Academy of Sciences of the United States of America, Jeanne, J. M., Sharpee, T. O., & Gentner, T. Q. (2008). Song recognition learning drives 103(4), 1088–1093. experience-dependent representations in the auditory forebrain region CLM. Paper Pinaud, R., & Terleph, T. A. (2008). A songbird forebrain area potentially involved in presented at the society for neuroscience. auditory discrimination and memory formation. Journal of Biosciences, 33(1), Jones, M. W., Errington, M. L., French, P. J., Fine, A., Bliss, T. V., Garel, S., et al. (2001). 145–155. A requirement for the immediate early gene Zif268 in the expression of late LTP Pinaud, R., Terleph, T. A., Tremere, L. A., Phan, M. L., Dagostin, A. A., Leão, R. M., et al. and long-term memories. Nature Neuroscience, 4(3), 289–296. (2008). Inhibitory network interactions shape the auditory processing of Karten, H. J. (1968). The ascending auditory pathway in the pigeon (Columba liva). II. natural communication signals in the songbird auditory forebrain. Journal of Telencephalic projections of the nucleus ovoidalis thalami. Brain Research, 11, Neurophysiology, 100(1), 441–455. 134–153. Pinaud, R., Velho, T. A. F., Jeong, J. K., Tremere, L. A., Leão, R. M., von Gersdorff, H., Karten, H. J. (1969). The organization of the avian telencephalon and some et al. (2004). GABAergic neurons participate in the brain’s response to birdsong speculations on the phylogeny of the amniote telencephalon. In J. M. Petras & C. auditory stimulation. European Journal of Neuroscience, 20(5), 1318–1330. R. Noback (Eds.). Comparative and evolutionary aspects of the vertebrate central Remez, R. E., Fellowes, J. M., & Rubin, P. E. (1997). Talker identification based on nervous system (Vol. 167, pp. 164–179). New York: The New York Academy of phonetic information. Journal of Experimental Psychology – Human Perception Science. and Performance, 23(3), 651–666. Kay, L. M., & Laurent, G. (1999). Odor- and context-dependent modulation of mitral Ribeiro, S., Cecchi, G. A., Magnasco, M. O., & Mello, C. V. (1998). Toward a song code: cell activity in behaving rats. Nature Neuroscience, 2(11), 1003–1009. Evidence for a syllabic representation in the canary brain. Neuron, 21(2), Kilgard, M. (2003). Cholinergic modulation of skill learning and plasticity. Neuron, 359–371. 38(5), 678–680. Rübsamen, R., & Dörrscheidt, G. J. (1986). Tonotopic organization of the auditory Klump, G. M., & Langemann, U. (1995). Comodulation masking release in a songbird. forebrain in a songbird, the European starling. Journal of Comparative Physiology, Hearing Research, 87(1–2), 157–164. 158, 639–646. Klump, G. M., & Maier, E. H. (1989). Gap detection in the starling (Sturnus vulgaris). Sachs, M. B., & Sinnott, J. M. (1978). Responses to tones of single cells in nucleus Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and magnocellularis and nucleus angularis of the redwing blackbird (Agelaius Behavioral Physiology, 164(4), 531–538. phoeniceus). Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, Kroodsma, D. E., & Miller, E. H. (Eds.). (1996). Ecology and evolution of acoustic and Behavioral Physiology, 126(4), 347–361. communication in birds. Ithaca: Comstock/Cornell. Saini, K., & Leppelsack, H. (1977). Neuronal arrangement in the auditory field L of Kruse, A. A., Stripling, R., & Clayton, D. F. (2004). Context-specific habituation of the the neostriatum of the starling. Cell and Tissue Research, 176(3), 309–316. ZENK gene response to song in adult zebra finches. Neurobiology of Learning and Saini, K. D., & Leppelsack, H. J. (1981). Cell types of the auditory caudomedial Memory, 82(2), 99–108. neostriatum of the starling. Journal of Comparative Neurology, 198, 209–230. Langemann, U., & Klump, G. M. (1992). Frequency discrimination in the European Searcy, W. A. (1992). Song repertoire and mate choice in birds. American Zoologist, starling (Sturnus vulgaris): A comparison of different measures. Hearing 32, 71–80. Research, 63(1–2), 43–51. Seeba, F., & Klump, G. M. (2009). Stimulus familiarity affects perceptual restoration Langemann, U., Klump, G. M., & Dooling, R. J. (1995). Critical bands and critical-ratio in the European starling (Sturnus vulgaris). PLoS One, 4(6), e5974. bandwidth in the European starling. Hearing Research, 84(1–2), 167–176. Sen, K., Theunissen, F. E., & Doupe, A. J. (2001). Feature analysis of natural sounds Leppelsack, H. J., & Vogt, M. (1976). Responses of auditory neurons in the forebrain in the songbird auditory forebrain. Journal of Neurophysiology, 86(3), of a songbird to stimulation with species-specific sounds. Journal of Comparative 1445–1458. Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 107(3), Shaevitz, S. S., & Theunissen, F. E. (2007). Functional connectivity between auditory 263–274. areas field L and CLM and song system nucleus HVC in anesthetized zebra Leppelsack, H.-J., & Schwartzkopff, J. (1972). Properties of acoustic neurons in the finches. Journal of Neurophysiology, 98(5), 2747–2764. caudal neostriatum of birds. Journal of Comparative Physiology, 80, 137–140. Sinnott, J. M., Sachs, M. B., & Hienz, R. D. (1980). Aspects of frequency discrimination MacDougall-Shackleton, S. A., Hulse, S. H., Gentner, T. Q., & White, W. (1998). in passerine birds and pigeons. Journal of Comparative and Physiological Auditory scene analysis by European starlings (Sturnus vulgaris): Perceptual Psychology, 94(3), 401–415. segregation of tone sequences. Journal of the Acoustical Society of America, Sockman, K. W., Gentner, T. Q., & Ball, G. F. (2002). Recent experience modulates 103(6), 3581–3587. forebrain gene-expression in response to mate-choice cues in European Maier, E. H., & Klump, G. M. (1990). Auditory duration discrimination in the starlings. Proceedings of the Royal Society of London Series B – Biological European starling (Sturnus vulgaris). Journal of the Acoustical Society of America, Sciences, 269(1508), 2479–2485. 88(2), 616–621. Sockman, K. W., Gentner, T. Q., & Ball, G. F. (2005). Complementary neural systems Maney, D. L., MacDougall-Shackleton, E. A., MacDougall-Shackleton, S. A., Ball, G. F., for the experience-dependent integration of mate-choice cues in European & Hahn, T. P. (2003). Immediate early gene response to hearing song correlates starlings. Journal of Neurobiology, 62(1), 72–81. with receptive behavior and depends on dialect in a female songbird. Journal of Stoddard, P. K. (1996). Vocal recognition of neighbors by territorial passerines. In D. Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral E. Kroodsma & E. H. Miller (Eds.), Ecology and evolution of acoustic Physiology, 189(9), 667–674. communication in birds (pp. 356–374). Ithaca: Comstock/Cornell. Mello, C., Nottebohm, F., & Clayton, D. (1995). Repeated exposure to one song leads Stoddard, P. K., Beecher, M. D., Horning, C. L., & Campbell, S. E. (1991). Recognition of to a rapid and persistent decline in an immediate early gene’s response to that individual neighbors by song in the song sparrow, a species with song song in zebra finch telencephalon. Journal of Neuroscience, 15, 6919–6925. repertoires. Behavioral Ecology and Sociobiology, 29, 211–215. Mello, C. V., Velho, T. A., & Pinaud, R. (2004a). Song-induced gene expression: A Sturdy, C. B., Phillmore, L. S., & Weisman, R. G. (2000). Call-note discriminations in window on song auditory processing and perception. Annals of the New York black-capped chickadees (Poecile atricapillus). Journal of Comparative Psychology, Academy of Sciences, 1016, 263–281. 114(4), 357–364. Mello, C. V., Velho, T. A. F., & Pinaud, R. (2004b). Song-induced gene expression: A Suzuki, R., Buck, J. R., & Tyack, P. L. (2006). Information entropy of humpback whale window on song auditory processing and perception. ANYAS, 1016, 263–281. songs. JASA, 119(3), 1849–1866. 68 D.P. Knudsen, T.Q. Gentner / Brain & Language 115 (2010) 59–68

Terleph, T. A., Mello, C. V., & Vicario, D. S. (2006). Auditory topography and temporal Vates, G. E., Broome, B. M., Mello, C. V., & Nottebohm, F. (1996). Auditory pathways response dynamics of canary caudal telencephalon. Journal of Neurobiology, of caudal telencephalon and their relation to the song system of adult male 66(3), 281–292. zebra finches. Journal of Comparative Neurology, 366(4), 613–642. Theunissen, F. E., Amin, N., Shaevitz, S. S., Woolley, S. M., Fremouw, T., & Hauber, M. Warren, R. M. (1970). Perceptual restoration of missing speech sounds. Science, E. (2004). Song selectivity in the song system and in the auditory forebrain. 167(3917), 392–393. Annals of the New York Academy of Sciences, 1016, 222–245. Wild, J. M., Karten, H. J., & Frost, B. J. (1993). Connections of the auditory forebrain in Theunissen, F. E., & Doupe, A. J. (1998). Temporal and spectral sensitivity of complex the pigeon (Columba livia). Journal of Comparative Neurology, 337(1), 32–62. auditory neurons in the nucleus HVc of male zebra finches. Journal of Woolley, S. M. N., & Casseday, J. H. (2004). Response properties of single neurons in Neuroscience, 18(10), 3786–3802. the zebra finch auditory midbrain: Response patterns, frequency coding, Theunissen, F. E., Sen, K., & Doupe, A. J. (2000). Spectral–temporal receptive fields of intensity coding, and spike latencies. Journal of Neurophysiology, 91(1), 136–151. nonlinear auditory neurons obtained using natural sounds. Journal of Woolley, S. M. N., Fremouw, T. E., Hsu, A., & Theunissen, F. E. (2005). Tuning for Neuroscience, 20(6), 2315–2331. spectro-temporal modulations as a mechanism for auditory discrimination of Thompson, J. V., & Gentner, T. Q. (2008). Song recognition learning elicits differential natural sounds. Nature Neuroscience, 8(10), 1371–1379. suppression of neural responses to learned songs in NCM. Paper presented at the Woolley, S. M., Gill, P. R., Fremouw, T., & Theunissen, F. E. (2009). Functional groups society for neuroscience. in the avian auditory system. Journal of Neuroscience, 29(9), 2780–2793.